JUNE 2002 TEAGUE ET AL. 1621

Low-Frequency Current Observations in the /Tsushima Strait*

W. J. T EAGUE,G.A.JACOBS,H.T.PERKINS, AND J. W. BOOK Naval Research Laboratory, Stennis Space Center, Mississippi

K.-I. CHANG AND M.-S. SUK Korea Ocean Research and Development Institute, Seoul, Korea

(Manuscript received 13 March 2001, in ®nal form 21 September 2001)

ABSTRACT High resolution, continuous current measurements made in the Korea/Tsushima Strait between May 1999 and March 2000 are used to examine current variations having time periods longer than 2 days. Twelve bottom- mounted acoustic Doppler current pro®lers provide velocity pro®les along two sections: one section at the strait entrance southwest of and the second section at the strait exit northeast of Tsushima Island. Additional measurements are provided by single moorings located between Korea and Tsushima Island and just north of Cheju Island in Cheju Strait. The two sections contain markedly different mean ¯ow regimes. A high velocity current core exists at the southwestern section along the western slope of the strait for the entire recording period. The ¯ow directly downstream of Tsushima Island contains large variability, and the ¯ow is disrupted to such an extent by the island that a countercurrent commonly exists in the lee of the island. The northeastern section is marked by strong spatial variability and a large seasonal signal but in the mean consists of two localized intense ¯ows concentrated near the Korea and coasts. Peak nontidal currents exceed 70 cm sϪ1 while total currents exceed 120 cm s Ϫ1. The estimated mean transport calculated from the southwest line is 2.7 Sv (Sv ϵ 106 m3 sϪ1). EOF analyses indicate total transport variations in summer are due mainly to transport variations near the Korea coast. In winter, contributions to total transport variations are more uniformly distributed across the strait.

1. Introduction The western channel contains a closed topographic de- pression in its center, approximately 60 km long, 10 km Of the four straits that connect the Japan/East Sea wide, and maximum depth 200 m. Its position is ap- with adjacent seas (Korea, Tsugaru, Soya, and Tatar proximated in Fig. 1 by the 150 m closed contour west Straits), the Korea/Tsushima Strait provides the vast ma- and northwest of Tsushima Island. jority of in¯ow and thus is instrumental in controlling Long-term measurement of currents in the Korea/ the Japan/East Sea properties. The major in¯ux of sea- Tsushima Strait are very dif®cult to acquire due to the water entering the Japan/East Sea through this strait is intense level of ®shing and trawling (Kawatate et al. important in the transfer of dynamic, thermodynamic, 1988). Hence, such measurements are relatively scarce. biological, and passive tracer properties into the Japan/ Many short-term current measurements have shown East Sea. This strait is effectively a shallow, connecting strong currents in addition to strong tidal components channel between the broad, shallow East Sea (Mizuno et al. 1989; Egawa et al. 1993; Isobe et al. (depths of approximately 50 to 1000 m) and the much 1994; Katoh et al. 1996). The major current features in deeper Japan/East Sea (depths exceed 2000 m). Korea/ the strait are the in¯ow of the warm Tsushima Current Tsushima Strait is about 330 km long and 100 m in from the , the out¯ow into the Japan/ depth. The strait is about 160 km wide and is divided East Sea, and a counter¯ow on the east side of Tsushima by Tsushima Island into the East and West channels. Island (Miita and Ogawa 1984; Egawa et al. 1993). The current direction is generally northeastward along the * Naval Research Laboratory/Stennis Space Center Contribution channel, but southwestward countercurrents have been Number JA/7330/01/0067. observed (Katoh 1994; Isobe et al. 1994). Tidal currents with speeds approaching 50 cm sϪ1 embedded in total currents of nearly 100 cm sϪ1 have been reported in the Corresponding author address: Mr. W. J. Teague, Meso- and Fine- scale Ocean Physics Section, Naval Research Laboratory, Stennis strait by Isobe et al. (1994). Space Center, MS 39529-5004. Currents in the strait are generally thought to exhibit E-mail: [email protected] considerable seasonal variation in current velocity (Ega-

᭧ 2002 American Meteorological Society

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC 1622 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 32

FIG. 1. ADCP mooring locations, and bathymetry. Depths are in meters. Cheju Strait is located between Cheju-Do and Korea. The Korea/Tsushima Strait is located between Japan and Korea. Bathymetry is from the 1999 Digital Atlas for Neighboring Seas of Korean Peninsula (provided by Prof. Byung-Ho-Choi, Laboratory for Coastal and Ocean Dynamics Studies, Sungkyunkwan University). wa et al. 1993). Due to the scarcity of long-term direct channel but also is present during all seasons. Little current measurements, particularly in the winter season, seasonal variation is observed but largest current speeds this variability has been deduced from seasonal varia- are observed in October through December time period tions in water mass properties and in the sea level dif- with maximum currents in the near-surface layer in De- ferences among tidal stations along the Korean and Jap- cember and at middepth in November. anese coasts. Many of the direct current measurements The development of trawl resistant mounts for moor- are made using ship-mounted acoustic Doppler current ing instruments on the bottom makes long-term mea- pro®lers (ADCPs) where removing ship motions and surements possible in the strait. This technique was used tides are often a large source of error. Large seasonal in the to deploy ADCPs and pressure gauges variation in the currents (higher magnitudes in the sum- (Teague et al. 1998; Teague and Jacobs 2000). This mer than in the winter) have been deduced using the study analyzes data from 14 similar moorings in the geostrophic current calculation (Yi 1966) and the sea strait (Fig. 1). Most of these moorings were deployed level differences (Yi 1970; Kawabe 1982; Toba et al. for about ten months. Twelve of the moorings (N1±N6, 1982). According to these studies, currents are at a max- S1±S6) are located on sections nearly spanning the strait imum in summer±fall and at a minimum in winter± to the northeast and southwest of Tsushima Island. The spring in the western channel. However, similar studies other two moorings are located just north of Cheju Is- in the eastern channel have found little seasonal vari- land and between Korea and Tsushima Island. One pur- ability (Isobe et al. 1994). Egawa et al. (1993) analyzed pose of these measurements is to delineate the nontidal ship-mounted ADCP observations taken over a three- currents and thus gain a better understanding of the year period. They found that the main axis of the Tsush- dynamics in the strait. Using the ®rst ®ve months of ima Current exists in the channel west of Tsushima Is- these measurements, Perkins et al. (2000a) present a land for all seasons and that the current ¯owing east of short description of the current ¯ow, and Jacobs et al. Tsushima Island is slower than the current in the western (2001b) estimate transports of 2.5 and 2.9 Sv (Sv ϵ 106

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC JUNE 2002 TEAGUE ET AL. 1623 m3 sϪ1) through the northeastern and southwestern lines, 1995). The U.S. Naval Oceanographic Of®ce imple- respectively. Along the southwestern section, nontidal mented similar moorings in 1995 in the Yellow Sea ¯ows exceed 50 cm sϪ1, and along the northeastern (Teague et al. 1998). Within the Korea/Tsushima Strait, section nontidal ¯ows exceed 70 cm sϪ1. The two sec- 14 mooring packages were deployed in a combined ef- tions show markedly different ¯ow regimes. At the fort between the Naval Research Laboratory, the Naval southern entrance the ¯ow through the strait contains a Oceanographic Of®ce, and the Korea Ocean Research broad maximum at midchannel. The northeastern sec- and Development Institute (KORDI). Six moorings tion, is marked by strong spatial variability, but in the were deployed along each of two sections that span the mean consists of two streams, one on each side of the entire Korea/Tsushima Strait. One section is located strait. Between the two is a regime of highly variable northeast of Tsushima Island (mooring positions de- ¯ow with a weak mean, presumably due to the sepa- noted by N1±N6 in Fig. 1) and the other is located ration of the ¯ow by Tsushima Island. Tides are found southwest of Tsushima Island (denoted by S1±S6 in Fig. to signi®cantly contribute to the total current ¯ow and 1). The deployment times generally cover the period are described by Teague et al. (2001). Total currents from May 1999 through March 2000. Moorings along with tides can be twice as large as the currents without the two sections were recovered and redeployed in Oc- tides. Inertial currents dynamics within the strait, par- tober 1999. Addition moorings were deployed in the ticularly important during summer, are examined by Ja- strait between Tsushima Island and Korea (C1) from cobs et al. (2001a). October 1999 through March 2000 and in Cheju Strait Numerous studies have proposed different theories (K1) from March 1999 through December 1999. Geo- on the dynamics of the generation mechanism and sea- graphical positions, deployment periods, water depths, sonal variation of the Tsushima Current. The low fre- and ADCP sample levels are provided in Table 1. quency current variations are expected to be mainly re- Each deployed package consists of an ADCP and a lated to the sea level drop between the East China Sea wave/tide gauge (except at C1 and K1) housed in a and the Japan/East Sea and ultimately linked to the me- trawl-resistant bottom mount (TRBM). Nine of the de- ridional sea level difference between the in¯ow and out- ployments are based on a new type of TRBM known ¯ow ports (Ohshima 1994). Height differences across as Barnys after their barnacle-like shape (Perkins et al. the strait are due to local wind stress, remote wind stress 2000b). The Barnys were developed by SACLANT across the adjoining seas, and steric variations. In ad- Center in Italy in collaboration with the Naval Research dition, the large-scale circulation in the Paci®c Ocean Laboratory (NRL). The remaining TRBMs, at sites C1, in¯uences the sea level in the East China Sea and the N2, N3, and S5, are of a design developed at the Naval sea level at the out¯ow through , and Oceanographic Of®ce (Teague et al. 1998). The mooring hence the sea level within the Japan/East Sea (Nof at K1, deployed by KORDI, consists of a commercially 2000). Local wind forcing (Huh 1982) and local strat- available TRBM. In general, these types of mounts are i®cation (Sekine 1988) due to the intrusion of bottom shallow dome-shaped enclosures that rest upon the bot- cold water (Isobe 1997) are also important in the sea- tom. The exposed side of the dome is relatively smooth sonal variation of the Tsushima Current. The forcing in order to minimize snagging by ®shing nets and lines. and dynamics that lead to variations examined here are Scraps of nets as well as evidence of trawl scrapings complex. Within this study, we only address the char- were found on the moorings at instrument retrieval. acteristics of variations and nature of the low frequency There was no evidence of adverse impact on the data ¯ow. records due to ®shing activities. Current pro®les measured by bottom moored ADCPs The Barny mounts and the commercial mount are over long time periods are ideal for answering questions equipped with RD Instruments Workhorse ADCPs op- pertaining to the currents and tides that have never been erating at 300 KHz. The TRBMs from the Naval Ocean- adequately addressed due to the dif®culty in making ographic Of®ce are equipped with RD Instruments Nar- such measurements. Several error sources are mitigated rowband ADCPs operating at 150 KHz. Most of the by the measurements. Namely, spatially sparse mea- packages contain Sea-Bird Electronics Model 26 wave/ surements using single moorings provide poor coverage, tide gauges and all contained EdgeTech Model 8202 and ship ADCP sections can be biased by short time acoustic releases for location and recovery. The instru- period events. However, caution should be exercised in ments, protected within the mount, rest about 0.5 m interpreting the results since the data re¯ects variability above the ocean bottom. The ADCPs are set up to pro- only through the one deployment time. vide current pro®les with an accuracy of 1 cm sϪ1 over nearly the full water column. Vertical resolution was set to 4 m, except at S1, S2, and N2 where resolution was 2. Instrumentation set to 2 m. Pro®les of U (east±west) and V (north±south) A technique for mooring ADCPs on the bottom in components of velocity are recorded at 30-min intervals areas of high ®shing activity was recently developed. except at 15-min intervals at S5, N2, and N3. This method utilizes dome-shaped mounting pods, For analyses of the low-frequency current, tidal cur- which are highly trawl resistant (Cumbee and Foley rents and currents near the tidal semidiurnal and diurnal

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC 1624 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 32

TABLE 1. ADCP summary.

Mooring Lat (Њ) Lon (Њ) Start day End day Top bin (m) Bottom bin (m) Bin size (m) Water depth (m) S1 34.32 127.90 130 289 5 53 2 59 290 437 6 55 1 S2 34.13 128.12 129 289 7 83 2 89 290 438 7 83 2 S3 33.93 128.34 129 288 11 103 4 113 290 438 11 103 4 S4 33.74 128.56 128 287 9 97 4 107 288 438 9 97 4 S5 33.54 128.78 128 287 19 143 4 152 S6 33.35 129.00 128 286 9 105 4 115 288 438 9 105 4 C1 34.95 129.26 292 444 7 97 2 103 K1 33.64 126.52 62 356 6 114 4 124 N1 35.31 129.60 293 440 9 113 4 119 N2 35.20 129.67 125 282 25 137 2 142 285 441 24 132 4 N3 35.01 129.99 126 283 13 125 4 132 285 441 11 123 4 N4 34.84 130.21 126 283 13 117 4 127 285 441 13 117 4 N5 34.67 130.43 127 284 16 120 4 130 285 442 11 119 4 N6 34.50 130.65 127 284 12 108 4 118 286 442 12 108 4 frequencies are removed from the ADCP records by to near the bottom or can take the form of a subsurface using a low-pass ®lter with a 40-h cutoff frequency. jet. Current ¯ow near the middle portion of the north

Velocities are then rotated (42.5Њ) such that the Vr com- section is sometimes relatively weak toward the north- ponent is approximately along strait, normal to the east. However, there is often a counter¯ow toward the

ADCP section, and Ur is cross strait, approximately par- southwest that can extend from the surface to the bot- allel to the section. tom. Such a current con®guration, as on 9 October, has been depicted by Miita and Ogawa (1984) using geo- strophic current calculations and short-term current me- 3. Snapshots ter measurements. The deep countercurrent in the de- An overview of along-strait current velocities is pro- pression on the northwestern end of the section, such vided in snapshots shown in Figs. 2 and 3. Velocities as on 23 August, is associated with the movement of are interpolated along the northeast and southwest sec- cold water into the Korea/Tsushima Strait from the Ja- tion after tide removal to provide two-dimensional pic- pan/East Sea and has been observed previously (Lim tures or snapshots of the ¯ow using the optimal inter- 1973; Kaneko et al. 1991). polation described in Jacobs et al. (2001b). Snapshots of the velocities for components normal to each section 4. Velocity distributions (Vr) are shown for representative current con®gurations (approximately 10 days apart) in Fig. 2 for the south- A wide range of velocities are observed in the strait. western section. A high velocity core with peak veloc- Nontidal velocities can be in excess of 75 cm sϪ1 at N1, ities greater than 30 cm sϪ1 are observed along the west- and tides can greatly affect these velocities. Current ern bathymetric slope of the south section for the entire speed distributions within 10 cm sϪ1 velocity bins, with 10-month period. The high velocity core sometimes ex- tides and without tides for near-surface velocities, are tends to the bottom. Another high velocity core is in- shown in Fig. 4. When tides are included speed distri- frequently observed on the eastern bathymetric slope in butions generally appear more Gaussian, and speeds can the upper half of the water column, sometimes not ex- nearly double. For example, at C1 during the entire tending to the surface layers. Current ¯ow toward the October 1999 through March 2000 recording period, southwest is sometimes observed on the northeastern largest low-frequency currents are observed in the 70± and southwestern ends of the section near land bound- 80 cm sϪ1 velocity bin while largest total currents (with aries. tides) are observed in the 160±170 cm sϪ1 velocity bin. Snapshots of the velocities for the northeastern sec- Inclusion of tides signi®cantly affects the speed distri- tion are shown in Fig. 3. High velocity currents directed butions, particularly at the low and high speed ends of toward the northeast are frequently observed near both the distributions. Current speeds can reach 100 cm sϪ1 ends of the section. They can extend from the surface throughout the strait.

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC JUNE 2002 TEAGUE ET AL. 1625

FIG. 2. Snapshots of the along-strait velocities for representative current con®gurations after tide removal are shown for the south section. The vertical axis is depth and the horizontal axis is along the section and across the Korea/Tsushima Strait from west to east. Positive velocities are toward the northeast.

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC 1626 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 32

FIG. 3. As in Fig. 2 but for the north section.

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC JUNE 2002 TEAGUE ET AL. 1627 velocity bins after tides have been 1 Ϫ removed. Asterisks indicate the frequency of current observations that include tidal currents. . 4. Overall and monthly near-surface speed histograms. Vertical bars indicate the frequency of current observations occurring within 10 cm s IG F

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC 1628 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 32

FIG. 5. (a) Vector stick diagrams of current velocities for ADCPs along the northern section, and C1. The currents have been vertically averaged for each ADCP. Tides have been removed using a low-pass ®lter with a 40-h cutoff frequency. X-axis units are in days beginning in 1999. (b) As in (a) but for the southern section, including mooring K1.

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC JUNE 2002 TEAGUE ET AL. 1629

FIG. 6. Mean currents near the surface (top velocity bin) (a), for mid-depth currents (middle velocity bin) (b), for near-bottom currents (bottom velocity bin) (c), and for depth-averaged currents, (d) and their corresponding standard deviation ellipses after tide removal are shown for the 11-month period (May 1999± Oct 2000).

Low speeds (0±10 cm sϪ1 bin) for the low frequency cause much of the ¯ow and its associated variability is currents dominate at the ends of the southwestern sec- to the northeast, the Vr component re¯ects most of the tion (S1, S6) and often at the highly variable N3 and signal. The exception to this rule occurs at N3 and N4, N4 locations. Higher speed distributions occur in the located downstream of Tsushima Island in an area of middle of the southwestern section and at the ends of weak ¯ow and relatively isotropic variability. the northeastern section. Low frequency current speeds Vector stick diagrams of current velocities averaged are generally smallest in January through March and over depth and sub sampled at 6-h intervals are pre- highest in October and November. Over the entire re- sented in Fig. 5. Current ¯ow is predominately toward cording periods, speed distributions are similar between the northeast except at moorings N3, N4, and S6 where K1 and S2, C1 and N1, N3 and N4, and N5 and N6. direction is highly variable. However, isolated south- westward ¯ow events occur at each of the moorings. Current characteristics appear to vary over timescales 5. Vertically averaged currents ranging from several days to about a week. Maximum Current ¯ow in the interior of the strait generally velocities, exceeding 60 cm sϪ1, are observed at N1. follows the bathymetry contours (Lie et al. 1998). Be- Along the southwestern section (Fig. 5b), current ¯ow

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC 1630 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 32

FIG. 7. Monthly time series of mean current vectors and their corresponding standard deviation ellipses for near- surface (top velocity bin) (a), mid-depth (b), and near-bottom (c) levels.

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC JUNE 2002 TEAGUE ET AL. 1631

FIG.7(Continued) increases to a broad maximum at midchannel, with cur- each of the moorings for several depth levels. Mean rent speeds of about 50 cm sϪ1. Large, northeastward currents for the entire 11-month period at depth levels velocity components are dominant at S2, S3, and S4, near the surface, at middepth, near the bottom, and for with the largest velocities at S3. Velocities are weaker vertically averaged currents with standard deviation el- at S1 and S6. Velocities at K1 in Cheju Strait are stron- lipses are shown in Fig. 6. The center of the standard ger than those at S1 but a little weaker than those at deviation ellipse is at the tip of the arrowhead and re- S2. Sporadic countercurrents or southwest ¯ows are ob- ¯ects the area that is within one standard deviation of served along both ends of the southwestern section at the mean. Mean directions appear to closely follow the S1, S2, and S6. The strongest southwest ¯ow burst along bathymetry at the southwestern section with the possible the south section occurs near day 265 at S1, S2, and exception of S1 where local bathymetry effects may be K1, exceeds 20 cm sϪ1, and has a duration of over a signi®cant. Flow at S1 is northward toward the Korea week. This burst is associated with Super Typhoon coast. The strongest currents have means much larger Bart's passage. than their respective deviation ellipses. Therefore, the Along the northeastern section (Fig. 5a), strongest best determined depth-averaged currents are at S3, C1, northeast ¯ows (Vr) are observed at the ends of the and N1. The strongest mean currents occur on the west- section at N1 and N2, and at N5 and N6. Currents are ern side of the strait. Mean current directions along the much weaker near midstrait at N3 and N4 and reverse northeastern section ¯ow to the northeast except for the direction frequently. Flow reversals due to the effects weaker currents observed at N3 and N4 and the near- of Bart are apparent at N2, N3, and N4. All of the bottom currents at N1 and N2, which are directed toward moorings on the northeastern section exhibit at least Korea. Largest variability, indicated by the major axes some southwestward ¯ow into the strait from the Japan/ of the standard deviation ellipses, generally appears in East Sea. Strong northeastward ¯ow at C1 appears sim- the direction of the mean ¯ows. ilar to the ¯ow at N1. The mean current vectors and their standard deviation ellipses are shown for individual months for near-sur- face, middepth, and near-bottom levels in Figure 7. 6. Monthly variability These vectors are similar to those shown in Fig. 6 but An overall picture of the current ®eld can be gained are not on a geographical grid for illustrative purposes. from mapping the currents averaged over the year at Stronger and more persistent ¯ows are observed in the

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC 1632 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 32

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC JUNE 2002 TEAGUE ET AL. 1633 ) (approximately parallel to the north and south sections). The r U ) (approximately normal to the south and north sections). r V zero velocity is indicated by the light vertical line. (b) As in (a) but for the along-strait velocity ( . 8. Monthly mean velocity pro®les (heavy line) and standard deviation (horizontal bars) for (a) the across-strait velocity ( IG F

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC 1634 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 32 interior of the strait along the southwestern section. Cur- winter and spring that is more pronounced in the near rents tend to be more variable and weaker near the surface. Variability in the current is such that a southerly coasts. Very small near-bottom ¯ows are observed at ¯ow is commonly observed during all months of the K1 and S6. A southwestern mean ¯ow in all three layers year. is found for February at S1. Overall variability is generally larger for the along- Along the northeastern section largest monthly mean strait component along the northeastern line. As ex- currents are found near Korea (N1 and N2). Mean cur- pected, mean velocities are much larger for the along- rents at N3 and N4 tend to be much weaker and more strait component. Currents become more nearly baro- variable in direction than at any other location. Occur- tropic during the winter months. rences of southwestward currents are common. Else- where along the northeastern section currents are almost 7. Transport always toward the northeast and show a more pro- nounced seasonality with stronger ¯ows in summer and Transport variations across the strait are computed fall. Near-bottom mean southwestward currents into the (Fig. 9) by vertically integrating the interpolated veloc- strait are only observed at N1 in January, opposite to ity sections as computed by Jacobs et al. (2001b). The the general northeastward ¯ow. high velocity core, centered at about 120 km off Korea, A seasonal signal consisting of larger mean velocities is apparent in the southwestern section. Southwestward during the summer and fall months is apparent in the ¯ow reversals are apparent at about 80 km off Korea near-surface layers. Variability is also correspondingly along the northeastern section and near the ends of the larger. Seasonality appears to be more pronounced along southwestern section. Seasonality is most pronounced the northeastern section than along the southwestern at the end of the northeastern section closest to Korea. section. Seasonality is reduced below the surface layer, Transport is higher in the channel west of Tsushima and seemingly more so along the southwestern section Island in October and November than in the eastern than along the northeastern section. Although mean cur- channel. Transport estimates for both sections are pre- rents are small at S6, a striking seasonality is evident sented as a function of time in Fig. 10. Transports pre- in the near-surface layer with countercurrents observed viously estimated by Jacobs et al. (2001a,b) for May from October to February. Along the northeastern line, through October 1999 are 2.9 Sv through the southern largest near-bottom mean velocities occur near the ends section and 2.5 Sv through the northern section with of the line (N1, N2, and N6) during spring months. The expected errors of about 0.5 Sv. The time-averaged large mean ¯ow in October at N1 is based only on about transports for May 1999 through March 2000 are 2.7 one week of data. Larger standard deviation ellipses Sv across the southwestern section and 2.3 Sv across occurring in September are likely associated with the the northeastern section, with standard deviations of passage of Typhoon Bart. about 0.9 for both lines.

Mean velocity pro®les for the Ur (cross-strait) and Vr The transport estimate for the northeastern section is (along-strait) components with standard deviations are 0.4 Sv smaller than the estimate for the southwestern shown in Figs. 8a and 8b. Larger velocities are clearly section. There are several reasons that may lead to trans- in the along-strait direction and near the surface. Largest port not observed through the northeastern section. Typ- monthly mean velocity, exceeding 60 cm sϪ1, is found ical scales observed are between 15 and 30 km. How- at N1 in November. Mean velocities exceeded 70 cm ever, narrow currents not resolvable by the moorings sϪ1 over the single week of data recorded at N1 in Oc- along the section could exist. High-resolution snapshots tober. Mean velocities of about 55 cm sϪ1 are observed of currents across the strait obtained by a ship-mounted at C1, located between Tsushima Island and Korea. ADCP (Isobe et al. 1994) indicate currents more narrow The mean pro®les generally re¯ect current ¯ow than the mooring spacing do occur. In addition, high through the strait from the Yellow and East China Seas velocity currents are also known to exist near the Korea and into the Japan/East Sea. Pro®les of Vr at C1 appear Coast that decrease away from the coast (Miita and Oga- to mirror pro®les observed at N1. Flow at K1 is qual- wa 1984; Egawa et al. 1993). These currents are not itatively similar to the ¯ow at S2. There are a number resolved by the northeastern section. Thus the north- of cases (at S1, S6, N1, N3, and N4) where the mean eastern section would underestimate the transport if ei- velocities re¯ect a counter¯ow. The counter¯ow occurs ther an intensi®ed coastal current or narrow currents throughout the mean pro®le at S1 in February and at between moorings exist. Finally, it is possible that the S6 in December. velocity ®eld becomes surface intensi®ed. The mea- The countercurrent that bifurcates from the eastern surements do not resolve the near-surface layer. Thus, side of the Tsushima Current along the southeastern end surface intensi®cation could also play a role in the of the section, near the west coast of the Goto Islands, smaller observed transport along the northeastern sec- has also been observed by Katoh et al. (1996) using tion. ship mounted ADCP instrumentation. The frequency of Short-term transport variations appear to occur on this countercurrent was unknown. The measurements timescales of three days to one week. Long-term trans- here at S6 show a monthly mean southward ¯ow in the port variations seem to steadily increase from about 1

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC JUNE 2002 TEAGUE ET AL. 1635

FIG. 9. Transport variability is indicated by the vertically integrated velocity as a function of distance (km) from Korea and time for the northeastern section (top) and southwestern section (bottom).

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC 1636 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 32

FIG. 10. Total transport estimates for the southwestern section (blue) and northeastern section (red) are shown as a function of time. The corresponding time-averaged transports are 2.7 Sv and 2.3 Sv. The rms transport error estimate is shown in the lower left by the con®dence intervals. These error estimates are time-invariant.

Sv in December to about 4 Sv in October, and then cies. The peaks at 1 and 2 cpd are due to tides (Teague begin to decrease rapidly in November. Monthly mean et al. 2001). transports are given in Table 2. Maximum monthly mean The peaks at about 1.1 cpd, present only during the transports are observed in October and minimum trans- summer for both sections, are associated with inertial ports in January. currents. There is a strong inertial oscillation response to wind stress in summer and a much weaker response in winter, even though wind events are comparable in 8. Spectra both seasons. Inertial oscillations within the Korea/ Energy spectra are computed from near-surface ve- Tsushima Strait indicate a vertical velocity structure in locity measurements for the northeast and southwest which the velocity reverses direction near the depth of sections (Fig. 11). Spectral estimates from 30-day seg- the mixed layer in summer. Strong summer strati®cation ments of velocity data sampled at 1-h intervals are av- prevents wind stress momentum ¯ux from mixing eraged for all moorings along each section. Spectra for downward and allows for a surface ¯ow toward the coasts with a subsurface return counter¯ow. The lack the along-strait velocity (Vr) are shown for each section for summer (May±mid October) and winter (mid Oc- of strati®cation in winter does not allow for such de- tober±March). Overall energy levels are higher along velopment of inertial currents. Inertial currents in the the northeast section than along the southwest section Korea/Tsushima Strait are fully described by Jacobs et for periods longer than 2 days (0.5 cpd). Energy levels al. (2001a). are higher along both sections during winter for periods from about two days to a week. The higher energies are 9. EOF analysis likely associated with more energetic wind events dur- ing winter. Wind pulses that have durations of about a The variance ellipses (Fig. 7) indicate the primary week are commonly observed in this region. Marginally directions of velocity variability for individual depth signi®cant peaks in the spectra for the northeast section levels for each ADCP. However, these ellipses alone do occur for both summer and winter for frequencies as- not provide an indication of how velocity variations may sociated with periods between 6 and 7 days. Peaks also be related between different depths and between moor- occur along the southwest section near these frequen- ings. Jacobs et al. (2001b) provide the horizontal and

TABLE 2. Monthly mean transport (Sv) for 1999±2000.

May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Avg ␴ N line 1.9 2.1 2.3 2.9 2.8 3.2 3.3 1.9 1.6 1.6 2.3 2.3 0.62 S line 2.7 2.8 2.8 3.0 3.1 3.5 3.1 1.8 1.7 2.2 2.8 2.7 0.55

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC JUNE 2002 TEAGUE ET AL. 1637

FIG. 11. Average energy spectra for moorings along the northeast (a) and southwest (b) sections for summer (solid line) and winter (dashed line). Con®dence levels indicated by the vertical bars correspond to the 95% signi®cance level. vertical decorrelation scales at each velocity level for The analysis is carried out separately for the observa- the ADCPs along the northeast and southwest sections. tions in summer (May through August) and winter (No- The horizontal length scales vary from 20 to 40 km with vember through March) since the wind forcing and a slight decrease at the thermocline depth. Vertical scales ocean strati®cation are signi®cantly different during vary from 10 m near the surface to 20 m near the bottom. these two time periods. September is excluded from the The spatial separation of the moorings is slightly less analysis because of an anomalously large event on day than the observed decorrelation scales. 260. This event is the ocean response to Super Typhoon To examine the main correlation patterns within the Bart that passed from the southwest across Kyushu. This ADCP datasets (northeast and southwest sections, only), single event can bias the covariance statistics and result the observations are decomposed by an empirical or- in a biased view of the typical correlations between thogonal function (EOF) analysis (Preisendorfer 1988). ADCPs. Each time series of u and ␷ is detrended over

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC 1638 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 32 the separate time periods to examine the interrelations that the band of variability is an island-induced effect. at timescales on the order of 1 month or less. It is well established that ¯ow from the strait eventually Only the ®rst two EOFs for winter and summer are takes two divergent paths in the Japan/East Sea. One examined. The ®rst summer EOF (Fig. 12a) indicates branch, the East Korean Warm Current, ¯ows northward most of the velocity variability near the Korean coast along the eastern coast of Korea. The second branch at moorings S1, S2, N2, and N3. The velocity variations (the Nearshore Branch) ¯ows eastward along the north- are to the northeast (when the time series amplitude is ern coast of Japan. However, the point at which these positive) and southwest (when the time series amplitude two currents bifurcate is not well established (Kaneko is negative), and the variations are surface intensi®ed. et al. 1991; Katoh et al. 1996). We presume that the two The time series indicates timescales of 5 to 7 days. The streams apparent at our northern mooring line are an overall impression from this ®rst mode is that the sum- expression of this separation at Tsushima Island. Pro®le mer transport variations in the strait are primarily due variability in Ur at N1 (Fig. 8a) points toward a north- to strengthening and weakening of the ¯ow on the Korea ward ¯ow up the coast of Korea. However, once this side of the strait. This ®rst summer EOF mode explains ¯ow has passed the steep bathymetric drop-off at the 18.8% of all the variability at all depths of all the ADCPs entrance to the Japan/East Sea, it is expected that a located on the northeast and southwest sections. The further bifurcation occurs and that two branches become second summer EOF mode (Fig. 12b) explains 15.8% dynamically separated (Katoh 1994). The branches are of the variability and is composed mainly of variability constrained to follow bathymetric contours. Another along the northern mooring line. This indicates a shift- possible ¯ow scenario suggested by Kim and Legeckis ing of the strait out¯ow between the Nearshore Branch (1986) is the occasional absence of the East Korean and the East Korea Warm Current. Warm Current. Instead, the ¯ow through the western The winter EOF decomposition (Fig. 12b) indicates channel of the Korea/Tsushima Strait forms an offshore more barotropic variability compared to summer. This branch along the Japanese coast following the conti- would be expected since the winter density structure is nental slope, while the Nearshore Branch on the con- nearly constant over depth while the summer density tinental shelf is fed by the ¯ow through the eastern structure contains a sharp pycnocline at depth 30 to 40 channel (Hase et al. 1999). The absence of the East m. The ®rst EOF mode indicates variability in the along- Korean Warm Current is attributed to the development strait direction that is relatively constant with depth. The of the southward movement of Korea/Tsushima Strait directions of the vectors across most of the south moor- bottom cold water (Cho and Kim 2000). ing line are similar. The north mooring line also indi- Interannual variations are unknown and could cause cates variations that are relatively constant over depth, signi®cant departures from this one measurement of the and the velocity variations near the Korea and Japan seasonal ¯ow. Seasonality in the currents along the coasts are in the same direction as those along the south southwestern section is weak. Near-surface layer cur- line. However, at N3 and N4 in the center of the north rents along the northeastern section are larger and usu- line the velocity variations are small and nearly in op- ally more highly variable in the summer and fall months posite direction to the other ADCPs. The implication is (Fig. 7a). Flows at middepths are smaller and less var- that the winter transport variations are due to homo- iable but similar to the near-surface ¯ows along both geneous currents across the strait except in the lee of sections (Fig. 7b). Very little seasonality is observed Tsushima Island. The second winter EOF (Fig. 12b) near the current core at middepth at S3. Flows near the contains a large divergence on the north line between bottom are smaller but similar to those at middepth (Fig. N1 and N2 and a large convergence between N3 and 7c) except at N1, N2, and N6 where the larger velocities N5. The ¯ow along the south line does not seem to play are found during the spring season. A more concentrated a large role in this variability. out¯ow along the northeastern section could magnify smaller variability in the main velocity core observed along the southwestern section. EOF analyses indicate 10. Summary and conclusions that the seasonality may be driven by current variability A persistent main velocity core of along-strait current on the Korea side of the strait. The ®rst two EOF modes ¯ow through the southwestern Korea/Tsushima Strait explain about 35% of the variability. lies along the bathymetric slope on the western side of The downstream transformation of the Tsushima Cur- the strait. The single current core apparent in the south- rent due to the con¯uence of Cheju Strait and East China western section separates into two branches near the Sea waters and the subsequent bifurcation of current Japan and Korea coasts as it splits around Tsushima into the eastern and western channels of the strait is still Island before reaching the northeastern section. In the poorly understood. The ¯ow through Cheju Strait passes lee of Tsushima Island, the two branches are separated mostly through the western channel of the Korea/Tsush- by a region (sites N3 and N4) where currents are variable ima Strait (Lie et al. 2000), and constitutes about 30% and lack a well-de®ned mean. Based only on these data, of the total transport through the western channel of the we conclude that the single current core found upstream Korea/Tsushima Strait (Chang et al. 2000). Through is divided by its passage around Tsushima Island and combined analyses of hydrographic and ship-mounted

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC JUNE 2002 TEAGUE ET AL. 1639

FIG. 12. EOF analyses of the moorings along both the northeast and southwest sections for (a) Mode 1 summer, (b) Mode 2 summer, (c) Mode 1 winter, and (d) Mode 2 winter. North is up along the y axis and east is to the right along the x axis. The vectors in the upper panels are scaled in time by the time series in the lower panels, and reverse direction on sign change by the time series. The vector velocity scale is indicated by the bar at the top of the ®gure. Horizontal bars at the bottom of each pro®le correspond to bottom depths.

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC 1640 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 32

ADCP measurements Katoh et al. (1996) concluded that current in the Tsushima Strait deduced from ADCP data of ship- the substantial bifurcation of the Tsushima Current oc- of-opportunity. J. Oceanogr., 49, 39±50. Hase, H., J.-H. Yoon, and W. Koterayama, 1999: The current structure curs in the vicinity of Tsushima Island. The measure- of the Tsushima Warm Current along the Japanese coast. J. ments here cannot determine where the current bifur- Oceanogr., 55, 217±235. cation occurs. If the ¯ow follows the bathymetry (Fig. Huh, O. K., 1982: Spring season ¯ow of the Tsushima Current and 1), then ¯ow through S1±S4 would pass west of Tsush- its separation from the KuroshioÐSatellite evidence. J. Geo- ima Island. However, the mean ¯ow directions suggest phys. Res., 87, 9687±9693. Isobe, A., 1997: The determinant of the volume transport distribution that the ¯ow through S1, S2, and S3 continues on west of the Tsushima Warm Current around the Tsushima/Korea of Tsushima Island while the ¯ow at S4, S5, and S6 Straits. Cont. Shelf Res., 17, 319±336. continues on east of Tsushima Island (Fig. 6). Nontidal ÐÐ, S. Tawara, A. Kaneko, and M. Kawano, 1994: Seasonal var- velocity distributions at S3 are more similar to the dis- iability in the Tsushima Warm Current, Tsushima±. Cont. Shelf Res., 14, 23±35. tributions at C1 than the velocity distributions at S4 (Fig. Jacobs, G. A., J. W. Book, H. T. Perkins, and W. J. Teague, 2001a: 4). These velocity distributions also suggest a split be- Inertial oscillations in the Korea Strait. J. Geophys. Res., 106 tween S3 and S4. The monthly mean ¯ow at S4 is always (C11), 26 943±26 957. more eastward than the mean current direction at S3. ÐÐ, H. T. Perkins, W. J. Teague, and P. J. Hogan, 2001b: Summer Such a ¯ow pattern would correspond to a larger trans- transport through the Korea±Tsushima Strait. J. Geophys. Res., 106, 6917±6929. port on the western side of Tsushima Island and is con- Kaneko, A., S.-K. Byun, S.-D. Chang, and M. Takahashi, 1991: An sistent with transports reported by Kaneko et al. (1991) observation of sectional velocity structures and transport of the and Katoh et al. (1996). Measurements here show that Tsushima Current across the Korea Strait. Oceanography of the transport is larger in the western channel for Sep- Asian Marginal Seas, K. Takano, Ed., Elsevier, 179±195. tember, October, and November. Katoh, O., 1994: Structure of the Tsushima Current in the south- western Japan Sea. J. Oceanogr., 50, 317±338. The strongest currents are observed off the coast of ÐÐ, K. Teshima, K. Kubota, and K. Tsukiyama, 1996: Downstream Korea and are approximately 100 cm sϪ1 but are nearly transition of the Tsushima Current west of Kyushu in summer. 170 cm sϪ1 if tidal current contributions are included. J. Oceanogr., 52, 93±108. The seasonal ¯ow pattern in the near-surface layer is Kawabe, M., 1982: Branching of the Tsushima Current in the Japan more pronounced northeast of Tsushima Island than Sea, Part I. Data analysis. J. Oceanogr. Soc. Japan, 38, 95±107. Kawatate, K., T. Miita, Y. Ouchi, and S. Mizuno, 1988: A report on southwest of Tsushima Island. The enhanced seasonal failures of current meter moorings set east of Tsushima Island signal may be due to more concentrated out¯ows near from 1983 to 1987. Progress in Oceanography, Vol. 21, Per- the coasts of Korea and Japan, where most of the current gamon, 319±327. ¯ow into the Japan/East Sea appears to occur. Seasonal Kim, K., and R. Legeckis, 1986: Branching of the Tsushima Current patterns become less pronounced at depth. Southwest- in 1981±1983. Progress in Oceanography, Vol. 17, Pergamon, 256±276. ward ¯ow events are not uncommon near the coasts Lie, H.-J., C.-H. Cho, and J.-H. Lee, 1998: Separation of the Kuroshio along the southwestern section and near the middle of water and its penetration onto the continental shelf west of Kyu- the strait along the northeastern section. Mean transport shu. J. Geophys. Res., 103, 2963±2976. through the strait, observed over the 10 months, is 2.7 ÐÐ, ÐÐ, ÐÐ, S. Lee, and Y. Tang, 2000: Seasonal variation of the Cheju Warm Current in the northern East China Sea. J. Sv with a standard deviation of 0.9 Sv. Transport is Oceanogr., 56, 197±211. minimum in January and increases to a maximum in Lim, D. B., 1973: The movement of the cold water in the Korea October. Strait. J. Ocean. Soc. Korea, 8, 46±52. Miita, T., and Y. Ogawa, 1984: Tsushima currents measured with Acknowledgments. This work was supported by the current meters and drifters. Ocean Hydrodynamics of the Japan and East China Seas, T. Ichiye, Ed., Elsevier, 67±76. Of®ce of Naval Research as part of the Basic Research Mizuno, S., K. Kawatate, T. Nagahama, and T. Miita, 1989: Mea- Projects ``Linkages of Asian Marginal Seas'' and ``Ja- surements of East Tsushima Current in winter and estimation of pan/East Sea DRI'' under Program Element 0601153N. its seasonal variability. J. Oceanogr. Soc. Japan, 45, 375±384. Chang and Suk were supported by grants from the KOR- Nof, D., 2000: Why much of the Atlantic circulation enters the Ca- DI's in-house project ``Marine Ecosystem Response to ribbean Sea and very little of the Paci®c circulation enters the . Progress in Oceanography, Vol. 45, Pergamon, Climate Variability in the East Sea and Tectonic Evo- 39±67. lution'' under Project Number PE00783. Ohshima, K. I., 1994: The ¯ow system in the Japan Sea caused by a sea-level difference through shallow straits. J. Geophys. Res., 99, 9925±9940. REFERENCES Perkins, H., W. J. Teague, G. A. Jacobs, K.-I. Chang, and M.-S. Suk, 2000a: Currents in Korea±Tsushima Strait during summer 1999. Chang, K.-I., M.-S. Suk, I.-C. Pang, and W. J. Teague, 2000: Ob- Geophys. Res. Lett., 27, 3033±3036. servations of the Cheju Current. J. Korean Soc. Oceanogr., 35, ÐÐ, F. de Strobel, and L. Gualdesi, 2000b: The Barny sentinel trawl- 129±152. resistant ADCP bottom mount: Design, testing, and application. Cho, Y.-K., and K. Kim, 2000: Branching mechanism of the Tsushima IEEE J. Oceanic Eng., 25, 430±436. Current in the Korea Strait. J. Phys. Oceanogr., 30, 2788±2797. Preisendorfer, R. W., 1988: Principal Component Analysis in Mete- Cumbee, S. A., and J. M. Foley Jr., 1995: Naval Oceanographic orology and Oceanography. Elsevier, 425 pp. Of®ce's ®rst operational use of trawl-resistant packages housing Sekine, Y., 1988: On the seasonal variation in in- and out¯ow volume acoustic Doppler pro®lers. Oceans, 2, 1372±1380. transport of the Japan Sea. Progress in Oceanography, Vol. 21, Egawa, T., Y. Nagata, and S. Sato, 1993: Seasonal variation of the Pergamon, 269±279.

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC JUNE 2002 TEAGUE ET AL. 1641

Teague, W. J., and G. A. Jacobs, 2000: Current observations on the Toba, Y., K. Tomizawa, Y. Kurasawa, and K. Hanawa, 1982: Seasonal development of the Yellow Sea Warm Current. J. Geophys. Res., and year-to-year variability of the Tsushima±Tsugaru Warm Cur- 105 (C2), 3401±3411. rent system with its possible cause. La Mer, 20, 41±51. ÐÐ, H. T. Perkins, Z. R. Hallock, and G. A. Jacobs, 1998: Current Yi, S. U., 1966: Seasonal and secular variations of the water volume and tide observations in the southern Yellow Sea. J. Geophys. transport across the Korea Strait. J. Oceanol. Soc. Korea, 1, 7±13. Res., 103, 27 783±27 793. ÐÐ, 1970: Variations of oceanic condition and mean sea level in ÐÐ, ÐÐ, G. A. Jacobs, and J. W. Book, 2001: Tide observations the Korea Strait. The Kuroshio: A Symposium on the Japan Cur- in the Korea-Tsushima Strait. Cont. Shelf Res., 21, 545±561. rent, J. C. Marr, Ed., East-West Center Press, 125±141.

Unauthenticated | Downloaded 09/29/21 09:14 PM UTC